Abstract
Aiming at the problems of low positioning accuracy and poor robustness of traditional odometers, this paper proposes a mileage positioning method based on subway feature recognition. First, use the YOLOv3 target detection method to identify the characteristics of the track fasteners, and obtain the position information of the characteristic points. Then, distance clustering and Lagrangian interpolation are used to complete the feature points that are not detected by deep learning. Finally, the initial mileage value is obtained according to the mileage constraint, and the speed filter method is used to correct the mileage result. In order to verify the reliability of the mileage positioning method, a mobile laser scanning experiment was carried out in Xi’an Metro. Experimental results show that this method has good positioning accuracy and stability. After Kalman speed filter preprocessing, the mileage positioning error is about 7cm, and the standard error is 5.05 × 10-3, which provides accurate mileage results for subway inspections.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have